modelscope / modelscope/ms-swift
monitor training efficiency performance metrics in logs and tensorboards
Open
Nobody has claimed this yet.
stale
- Dominant language
- Python
- Stars
- 15.7k
- Forks
- 1.7k
- Avg merge
- 1d 16h
- Merged PRs (30d)
- 136
Description
How do you guys generally monitor the key training performance metrics of the model during training?
For example, tgs, MFU, etc.
I feel that there can be a module that can support mainstream models and custom extensions
Contributor guide
First steps
- Read the whole issue, then the project's contributing guide.
- Comment on the issue to say you are picking it up — it saves two people doing the same work.
- Fork the repository and make your change on a branch.
- Open a pull request that references the issue number.
Research direction
The issue does not identify implementation files, entry points, or tests. Start by locating the training metric and logging integrations for mainstream models, then review how custom extensions could be supported. Done should include a defined module that records metrics such as tgs and MFU in logs and TensorBoard.
Written by the indexing model from the issue text.
Assessment
- Tech stack
- python
- Domain
- machine-learning, observability
- Issue type
- Feature
- Difficulty
- 5/5
- Estimated time
- Over a week
- Activity status
- Stale
- Clarity
- Needs clarification
- Newbie friendliness
- 25/100